GrowthandpropertiesofKLu(WO ,andnovelytterbium ... · 2007. 5. 15. · Laser & Photon. Rev. 1,...

Laser & Photon. Rev. 1, No. 2, 179–212 (2007) / DOI 10.1002/lpor.200710010 179

Abstract: High-quality crystals of monoclinic KLu(WO4)2,shortly KLuW, were grown with sizes sufficient for itscharacterization and substantial progress was achieved in thefield of spectroscopy and laser operation with Yb3+- andTm3+-doping. We review the growth methodology for bulkKLuW and epitaxial layers, its structural, thermo-mechanical,and optical properties, the Yb3+ and Tm3+ spectroscopy, andpresent laser results obtained in several operational regimesboth with Ti:sapphire and direct diode laser pumping usingInGaAs and AlGaAs diodes near 980 and 800 nm, respec-tively. The slope efficiencies with respect to the absorbedpump power achieved with continuous-wave (CW) bulk andepitaxial Yb:KLuW lasers under Ti:sapphire laser pumpingwere ≈ 57 and ≈ 66%, respectively. Output powers ashigh as 3.28 W were obtained with diode pumping in asimple two-mirror cavity where the slope efficiency withrespect to the incident pump power reached ≈ 78%. PassivelyQ-switched laser operation of bulk Yb:KLuW was realizedwith a Cr:YAG saturable absorber resulting in oscillation at≈ 1031 nm with a repetition rate of 28 kHz and simultaneousRaman conversion to ≈ 1138 nm with maximum energiesof 32.4 and 14.4 µJ, respectively. The corresponding pulsedurations were 1.41 and 0.71 ns. Passive mode-locking by asemiconductor saturable absorber mirror (SESAM) producedbandwidth-limited pulses with duration of 81 fs (1046 nm,95 MHz) and 114 fs (1030 nm, 101 MHz) for bulk and epitaxial

Projection of the KLu(WO4)2 structure parallel to theb crystallographic direction [010].

Yb:KLuW lasers, respectively. Slope efficiency as highas 69% with respect to the absorbed power and an outputpower of 4 W at 1950 nm were achieved with a diode-pumped Tm:KLuW laser. The slope efficiency reachedwith an epitaxial Tm:KLuW laser under Ti:sapphire laserpumping was 64 %. The tunability achieved with bulkand epitaxial Tm:KLuW lasers extended from 1800 to1987 nm and from 1894 to 2039 nm, respectively.

c© 2007 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Growth and properties of KLu(WO4)2, and novel ytterbiumand thulium lasers based on this monoclinic crystallinehostValentin Petrov 1,�, Maria Cinta Pujol 2, Xavier Mateos 1, Òscar Silvestre 2, Simon Rivier 1, Magdalena Aguiló 2, RosaMaria Solé 2, Junhai Liu 1, Uwe Griebner 1, and Francesc Dı́az 2

1 Max-Born-Institute for Nonlinear Optics and Ultrafast Spectroscopy, Max-Born-Str. 2A, 12489 Berlin, Germany2 Universitat Rovira i Virgili, Campus Sescelades, c/ Marcel·lı́ Domingo, s/n, 43007 Tarragona, Spain

Received: 10 April 2007, Accepted: 13 April 2007Published online: 3 May 2007

Key words: rare earth solid state lasers, monoclinic double tungstates, crystal growth from flux, epitaxial layers, ytterbium lasers,thulium lasers, diode-pumped lasers, ultrafast lasers

PACS: 42.55.Rz, 42.70.Hj, 81.10.Dn, 81.15.Np, 42.55.Xi, 42.60.Pk, 42.60.Fc, 42.60.Gd, 42.65.Dr

� Corresponding author: e-mail: [email protected]

c© 2007 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

180 V. Petrov et al.: Growth and properties of KLu(WO4)2

1. Introduction

The main advantages of the monoclinic potassium doubletungstates as laser hosts are the very high values of theabsorption and emission cross sections of the rare-earthdopants, partly due to the strong anisotropy of these biaxialcrystals, and the possibility to dope them with high concen-tration of the active ions without substantial fluorescencequenching. Information on the synthesis, crystal structureand some thermal properties of monoclinic KLu(WO4)2(KLuW), one of the optically inert compounds belong-ing to this class together with KY(WO4)2 (KYW) andKGd(WO4)2 (KGdW), appeared for the first time in 1969[1]. Many properties of KLuW like the thermal conductiv-ity, hardness, optical transparency, and refractive index arevery similar to those of the isostructural KYW [2]. Singlecrystals of KLuW were used to analyze the infrared andRaman spectra [3], and efficient high-order Stokes andanti-Stokes stimulated Raman scattering (SRS) in the vis-ible and near-infrared was observed for the two SRS-activemodes at 907 and 757 cm−1 [2].

Laser emission of Er:KLuW was demonstrated for the0.85, 1.74 and 2.81 µm transitions [4], the luminescenceproperties and laser operation of Ho:KLuW for the 2.08and 2.94 µm lines were also studied [5–7], but most of thespectroscopic and laser works were devoted to Nd:KLuW[7–10], where the 1.07 and 1.35 µm lines were investi-gated and CW generation at 1070.2 nm was achieved atroom temperature with diode pumping [10], see Table 1.Recently, some additional activities were focused on thegrowth, spectroscopy and diode-pumped laser operationof Nd:KLuW [11–13]. However, it is known from the ear-lier work that KNd(WO4)2 has a different structure, andNd-doping of KLuW is limited to about 3 at. % becausestresses and cracks occur [8,14].

The KLuW host is actually predestined for doping withYb because of the close ionic radii and masses of Yb andLu, and the close lattice parameters of the isostructuralKLuW and KYb(WO4)2 (KYbW) [25]. This allows highYb-doping levels with low defect formation probabilityand epitaxial growth of highly absorbing films with bestquality of the interface. Moreover, the Yb-dopant affectsonly weakly the thermal conductivity of the host. The nextmost suitable dopant for KLuW is Tm. In that case thetwo hosts KLuW and KYW have similar deviations fromthe isostructural KTm(WO4)2. Here we will review thegrowth, structure, and physical properties of KLuW, thespectroscopy of Yb3+ and Tm3+ in KLuW, and the laserresults obtained by us with bulk and epitaxial Yb:KLuWand Tm:KLuW crystals (see Table 1). Note that morerecently, KLuW was also doped with Sm [26], Dy [27],and Tb [28], as well as co-doped with Er:Yb, by the presentauthors and in [27], and with Yb:Tm [29], but lasing hasnot been demonstrated in such crystals, yet.

2. Crystal structure and growth of bulkand epitaxial KLuW

The structure of the low-temperature monoclinic phase ofKLuW was studied as early as 1968 but the lattice parame-ters determined by X-ray powder diffraction were reportedin the I2/c space group [1]. Subsequently, they were re-vised and refined in the C2/c space group, however, stillusing powdered crystals [14,25]. Our results derived fromsingle-crystal X-ray diffraction data [30] are included inTable 9. Similar results with single crystals were reportedalso in [31]. KLuW has the structure of KYW, as the otherpotassium double tungstates of the heavier lanthanidesstarting from KSm(WO4)2 (KSmW) for which the unitcell parameters and hence the interatomic distances tendto decrease with the Ln atomic number [1].

Lutetium is eightfold coordinated by oxygen atoms,forming a distorted square antiprism. The local site sym-metry of the Lu3+ cation is C2 (4e Wyckoff position), andsubstitution by dopant ions takes place in this unique site.The LuO8 polyhedra form a single zig-zag chain in the[101] direction sharing O—O edges [3.081(11) Å]. Thelength of these shared O—O edges increases along theKSmW-KLuW series due to the increasing positive chargeof the nucleus of the Ln-element. As a consequence, thestrength of one of the Lu—O bonds increases and the in-teratomic separation is decreased to 2.299(7) Å in compar-ison to the separation of 2.371(10) Å for Gd—O in KGdW[32]. The resulting degree of distortion of the LuO8 coor-dination polyhedra calculated for the KLuW host is morethan two times higher in comparison to KGdW [30].

The distances between Lu—Lu pairs are also impor-tant because they affect the energy transfer between thedopant ions. The Lu—Lu distance in the same LuO8 chainparallel to the [101] direction is 4.045(3) Å. Each suchlutetium polyhedra chain is surrounded by four equiva-lent chains, and the corresponding Lu—Lu distances are5.982(3) Å and 6.693(3) Å. This environment is similar tothat of KGdW where the corresponding Gd—Gd separa-tions amount to 4.070(2) Å, 6.057(2) Å, and 6.804(2) Å.

The coordination figure of the tungstate anion is a dis-torted octahedron, WO6, (W—O distances ranging from1.767(7) Å to 2.265(8) Å). The units W2O8, which are con-stituted by two distorted octahedra sharing O—O edges,form a characteristic double chain in the crystallographicc direction by sharing vertex O. The average W—O dis-tance gets shorter along the KSmW-KLuW series whichmeans decreasing distortion of the WO6 octahedra lead-ing to more compact and covalent WO6 groups. As in theother monoclinic double tungstates, the alkali cation K+ istwelve-coordinated by O ions, forming a distorted icosa-hedron. Figure 1 shows the LuO8 zig-zag chains alongthe [101] direction and the W2O8 double chains along the[001] direction.

The knowledge of the unit cell variation with dopingis important e.g. for epitaxial growth of doped layers onundoped substrates or for diffusion bonding of doped and

c© 2007 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.lpr-journal.org

Laser & Photon. Rev. 1, No. 2 (2007) 181

Table 1 Laser operation demonstrated with lanthanide (Ln) doped (activated) KLuW.

Ln3+ dopingƒ

[at. %]polarization◊ transition wavelength [nm] pump remarks* [Ref.] (year)

Nd 3 ⊥a 4F3/2→4I13/2 1348.2 Xe-lamp pulsed [7] (1979)Nd 2 ⊥a’, ⊥c 4F3/2→4I13/2 1353.3, 1355Nd 2 ⊥b 4F3/2→4I13/2 1355

Xe-lamp pulsed [8] (1983)

Nd 3 ⊥a 4F3/2→4I11/2 1071.4 Xe-lamp pulsed [7] (1979)Nd 2 ⊥a’ 4F3/2→4I11/2 1072.1Nd 2 ⊥b, ⊥c 4F3/2→4I11/2 1070.1, 1070.2

Xe-lamp 77 and 300 K, pulsed [8] (1983)

Nd ∼5 NA 4F3/2→4I11/2 1070.1 pyro-lamp quasi-CW [9] (1983)Nd 3 ⊥b 4F3/2→4I11/2 1070.2 diode laser quasi-CW, CW [10] (1992)Nd 3 ⊥b 4F3/2→4I11/2 1073 diode laser CW [13] (2005)Er 3, 5, 25 ⊥b, ⊥c 4I11/2→4I13/2 2809.2Er 3, 5 ⊥b 4S3/2→4I9/2 1738.3Er 5 ⊥b 4S3/2→4I13/2 ≈850

Xe-lamp pulsed [4] (1979)

Ho 3 ⊥c 5I7→5I8 2079 Xe-lamp 110 K, pulsed [6] (1981)Ho 3 ⊥b 5I6→5I7 2944.5 Xe-lamp pulsed [7] (1979)Yb 5, 10 //Nm 2F5/2→2F7/2 1033.3-1051.3 Ti:sapphire CW [15,16] (2004)Yb 5 //Nm

2F5/2→2F7/2 1041-1047 diode laser CW [15,16] (2004)Yb 5 //Nm 2F5/2→2F7/2 1039.5-1052.4 diode laser CW [17] (2005)Yb 5 //Np

2F5/2→2F7/2 1046 Ti:sapphire CW [18] (2005)Yb 5 //Nm

2F5/2→2F7/21st Stokes

1030.61137.6

diode laser Q-switched+ self-Raman laser

[17] (2005)

Yb 5 //Nm2F5/2→2F7/2 1046 Ti:sapphire M-L, fs [18] (2005)

Yb 5 //Np2F5/2→2F7/2 1043, 1049

1053Ti:sapphirediode laser

M-L, ps & fs,M-L, fs

[18] (2005)

Yb 10 //Nm2F5/2→2F7/2 1026-1040

1030Ti:sapphirediode laser

epitaxy, CW [19] (2005)

Yb 50 //Nm2F5/2→2F7/2 1032, 1046.1 Ti:sapphire epitaxy, quasi-CW, CW [20] (2006)

Yb 10 //Nm2F5/2→2F7/2 1030 Ti:sapphire epitaxy, M-L, ps & fs [21] (2005)

Tm 3 //Nm3F4→3H6 1943-1975 diode laser CW [22,23] (2006)

Tm 3, 5 //Nm3F4→3H6 1917-1951

1809-1983Ti:sapphire CW

CW-tunableTm 3 //Np

3F4→3H6 1907-19421800-1987

Ti:sapphire CWCW-tunable

[23] (2006)

Tm 5 //Nm3F4→3H6 1960-1967

1894-2039Ti:sapphire epitaxy, CW

CW-tunable[24] (2007)

f doping level in the solution; the data on the segregation coefficient of Nd in [11,13] is contradictory; ♦ the crystallographic a’axis (I2/c space group) is at ≈ 44.7◦ from the a-axis (C2/c space group); * M-L: mode-locked; if not mentioned, the temperatureis 300 K.

undoped KLuW. For Yb and Tm doping we studied thevariation of the unit cell parameters of KLuW by the X-ray powder diffraction method. The samples used weregrown with 10, 20, 50 and 100 at. % Yb doping and 3,5, 7.5, 10, and 20 at. % Tm doping in the solution butthe actual crystal composition, determined from electronmicroprobe analysis, was used to obtain the evolution ofthe unit cell parameters in Fig. 2 [30,33].

As could be expected, the monoclinic symmetry ismaintained in all cases because KYbW and KTmW havethe same structure as KLuW [1]. The unit cell parametersa, b and c increase and β remains basically constant when

the doping concentration of Yb and Tm increases. Thisincrease is also not unexpected because the ionic radii ofTm (0.994 Å) and Yb (0.985 Å) are larger than that of Lu(0.977 Å).

Polarized infrared and Raman spectra of KLuW weremeasured in the past and compared with the other isostruc-tural potassium double tungstates [3]. Due to the presenceof two non-equivalent structural groups in the primitivecell, the factor group analysis predicts a total of 72 vi-brations distributed among the following irreducible rep-resentations: 17Ag+19Bg+17Au+19Bu. The Ag and Bgmodes are Raman active, and the Au and Bu modes are

www.lpr-journal.org c© 2007 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim


Figure 1 Projection of the KLuW structure parallel to the bcrystallographic direction [010].

0.0 0.2 0.4 0.6 0.8 1.00.0

0.1

0.2

0.3

0.4

0.5 (a)Yb:KLuWabc

V

(L-L

0/L 0

) x

102

Yb content in crystal x

0.00 0.05 0.10 0.15 0.200.00

0.05

0.10

0.15

0.20

0.25

0.30 (b)Tm:KLuW

Tm content in crystal x

(L-L

0/L 0

) x

102

abc

V

Figure 2 Evolution of the unit cell parameters (L = a, b, c, β,)and the unit cell volume (L = V ) with the dopant concentrationx of KLu1−xYbxW (a) and KLu1−xTmxW (b).

infrared active. The modes can be further subdivided intoacoustic (T = 1Au + 2Bu), optical translational latticemodes (T ′ = 2Ag +4Bg +4Au+5Bu), optical librationallattice modes (L = 3Ag+3Bg) and optical internal modesfor the W2O8 units (int = 12Ag +12Bg +12Au +12Bu)[34,35].

Table 2 Vibrational frequencies for KLuW.

assignment energy [cm-1]

T’(W6+) / ? −− 303 νν 63

T’(W6+) / Bg 87

L(WO6)*δ(WOW) / Eg 112

? / Eu 130

γ(WOOW)*T’(K+) / Eu /031 νν − 147

γ(WOOW)*T’(K+) / Au 175

T’(Lu3+) / Eg 218

T’(K+) / Bg 236

δ(WOOW) / ? 289

γ (WOW) / −2ν 317δ(W – O) / ? +2ν 345δ(W – O) / ? +2ν 350δ(W – O) / ? 381

ν(W – O) / −4ν 406δ(WOW) / 04ν 450ν(WOOW) / −4ν 535ν(WOOW)*ν(W – O) / −3ν 686ν(WOOW)*ν(W – O) / 03ν 756ν(WOW)*ν(W – O) / +3ν 809ν(W – O) / ν1 908

0312 νν − 1063

ν : stretching mode, δ: bending mode, γ: out-of-plane mode,* coupling of the vibrations, ? unidentified

The values of the phonon energies we determined fromRaman spectra recorded with single crystals of KLuW,excitation by an Ar-laser at 514 nm, and backscatteringgeometry, are summarized in Table 2. The labelling of theassignment of the peaks is according to the more recentclassification in [34,35] and a simplified nomenclature de-rived from the scheelite structure [36]. Figure 3 shows aroom temperature Raman spectrum recorded in the fre-quency range 0 to 1200 cm−1. We used here designationsrelated to the optical indicatrix (see next section), as usualin spectroscopy. The phonons below 270 cm−1 can beattributed to external lattice modes associated with trans-lational motion of the cations of the structure (K+, Lu3+,and W6+) and rotational motion of WO6 groups in theunit cell. The bending modes appear in the 270–460 cm−1region, and the stretching modes in the 400–1000 cm−1region. The bands in the 430–750 cm−1 region, not ob-served for isolated WO4 tetrahedra, are related to doubleoxygen bridge vibrations activated when the coordinationnumber of tungsten increases from 4 to 6 [34].



200 400 600 800 1000

inte

nsity

[a.u

.]

energy [cm-1]

Figure 3 Raman spectrum of KLuW: p(gm)p̄ geometry.

The boldface numbers in Table 2 designate thestrongest Raman lines, interesting for SRS. The linewidthof the 756 and 908 cm−1 modes is 13.7 and 8.5 cm−1,respectively, somewhat less than the 16.6 and 12.5 cm−1reported for 5 at. % Yb:KLuW in [37] (note the differentframe in [37]).

A polymorphic transformation of KLuW just belowthe melting point of 1363 K was suggested but not re-solved in [1]. Thus, the low-temperature monoclinic phasecan be grown only from high-temperature solutions. Poly-morphism is typical for the potassium rare-earth doubletungstates, and for KLuW, orthorhombic (with layeredKY(MoO4)2 structure) and trigonal (with KAl(MoO4)2structure) modifications are also known [25,38,39]. Thetemperature of the first transition to the orthorhombicphase was given as 1293–1298 K [8,14] and that of thesecond transition – as just below the melting temperatureof 1363 K [1,14]. The results of our differential thermalanalysis are shown in Fig. 4 and included in Table 9.Whereas the slightly lower melting temperature seems tobe confirmed also by analogous measurements of 5 at. %Yb-doped KLuW [40], the two phase transitions seem notto be resolved in Fig. 4 and also in [40].

We grow KLuW by the top-seeded solution growth(TSSG) slow-cooling method with a cylindrical verticalfurnace as described elsewhere [41]. The composition isabout 12/88 mol % solute/solvent (K2W2O7). This is thestandard solvent used for KLuW [1], with the basic advan-tages being the absence of foreign ions, the low meltingtemperature, and the low viscosity during growth as a re-sult of the higher W content [41]; another possibility isK2WO4 [1,11,12] but it does not ensure good homogeni-sation of the solution and is more prone to evaporation.

The solubility curve of monoclinic KLuW in K2W2O7is shown in Fig. 5. The limits of the solubility curve aregiven by the properties of this binary system. At a solutecontent of ≈ 54 mol %, the phase that crystallises firstwhen decreasing the temperature should be the orthorhom-bic one. The lower limit of ≈ 5 mol % for the solute isrelated to the economical profitability of the growth pro-

400 600 800 1000 1200 1400

-0.6

-0.3

0.0

0.3

0.6

0.9

1326 K

1312 K

tem

pera

ture

diff

eren

ce [K

/mg]

temperature [K]

Figure 4 Differential thermal analysis of KLuW.

0 10 20 30 40 50 60 701050

1100

1150

1200

1250

1300

tem

pera

ture

[K]

KLuW [mol %]

Figure 5 Solubility curve of KLuW in K2W2O7.

cedure. Although the monoclinic phase of KLuW can begrown for solute content between 5 and 54 mol %, it isbetter to remain in the 5–15 mol % range where vari-ations of the temperature lead to only small changes inthe supersaturation. The growth temperature when usingsuch solutions is 1090–1175 K which means relativelylow economical cost of the growth procedure.

The knowledge of some physical properties of thegrowth solutions, such as density, surface tension, viscos-ity and thermal conductivity is of great importance foroptimizing the conditions for single crystal growth by theTSSG method. In order to determine the optimum condi-tions for the growth of single crystals of KLuW, the tem-perature dependence of the density, surface tension andviscosity was studied for a solution of KLuW/K2W2O7 =12/88 mol %.

The density measurements were performed by theArchimedean method based on two bobs with identicalsupports. The surface tension was obtained by the dip-ping cylinder method, based on the measurement of themaximum pull exerted on the lower open end of a Pt/Pt-10%Rh cylinder, just before the meniscus is broken. The



1100 1150 1200 12503.88

3.90

3.92

3.94

3.96

3.98

4.00

saturated solution (crystal growth range)

dens

ity [g

/cm

3 ]

temperature T [K]

Figure 6 Thermal evolution of the density of theKLuW/K2W2O7 solution for 12 mol % KLuW.

weight measurements, made with the same balance, wererecorded with a frequency of 1 Hz while the cruciblewas lowered at a rate of 1 cm/min. The viscosity mea-surements were performed by the damping method. Anopen-end cylinder of Pt/Pt-10%Rh was immersed in thesolution, suspended by a torsion wire. The oscillation at-tenuation was observed by means of a laser beam reflectedon a mirror located in the axial plane.

Figures 6 and 7 show the density, surface tension andviscosity of the solution versus temperature. As can beseen, the density and surface tension decrease linearlywith increasing temperature. The viscosity also decreaseswith the temperature having a similar behaviour.

Taking into account the solubility curve of KLuW inK2W2O7, the saturation temperature TS of the solutionwas estimated to be about 1145 K. All measurements wereperformed above this temperature in order to avoid crys-tallization although the aim was to determine the physicalproperties of the solution in the temperature range wherethe crystal growth takes place. The cooling interval forthe growth of bulk KLuW crystals is about 20 K. The ex-trapolation of the obtained curves into this range leads tothe following mean values for the density, surface tensionand viscosity, respectively: 3.98 g/cm3, 108.8 dyn/cm and38.8 cP.

The crystals are grown in Pt crucibles using K2CO3,WO3, and Ln2O3 (Ln = Lu, Yb, Tm) as starting materials,with 99.9% purity. A KLuW seed oriented along the b-crystallographic axis is attached to a Pt holder rotatingat 40 rpm. The temperature gradient in the solution isabout 1 K/cm in the vertical and radial directions withthe bottom and walls being hotter. The growth occurs as aresult of supersaturation when cooling below the saturationtemperature which depends on the flux composition andslightly increases with the doping level. The cooling rateduring growth varies between 0.1 and 0.3 K/h, dependingon the crystal size and the doping level. Finally, the growncrystals are removed slowly from the solution and cooledat a rate of 15–25 K/h, slightly above the surface.

1100 1150 1200 1250

100

102

104

106

108

110

saturated solution(crystal growth range)

surf

ace

tens

ion

[dyn

/cm

]

temperature T [K]

(a)

1100 1150 1200 125030

32

34

36

38

40

saturated solution (crystal growth range)vi

scos

ity [c

P]

temperature T [K]

(b)

Figure 7 Thermal evolution of the surface tension (a) and thedynamic viscosity (b) of the KLuW/K2W2O7 solution for 12mol % KLuW.

Figure 8 shows the distribution ofYb in a single crystalof Yb:KLuW. The segregation coefficient KYb or KTm,which is normally close to unity, ensures rather homoge-neous distribution of the doping ions, both Yb and Tm,

Table 3 Crystal growth data for Yb-doped KLuW. A: Yb substi-tution in solution relative to Lu [at. %], B: Cooling rate duringgrowth [K/h], C: Cooling interval [K], D: Crystal weight [g], E:Growth rate [10−4 g/h], G, H, and I: Crystal dimensions [mm]along the c, a*, and b crystallographic directions, respectively.

A B C D E G H I KYb0.5 0.1 18.5 4.3082 232 9.5 6.6 4.7 1.391.0 0.1 18.9 4.5719 242 12.2 11.8 7.8 1.37

1.5 0.1 18.8 4.2797 228 19.5 8.4 6.3 1.61

3 0.1 17.6 3.3310 189 10.0 10.9 9.3 1.455 0.1 17.2 3.4670 202 10.1 11.6 8.0 1.3710 0.2 20 2.7293 273 12.1 9.5 6.9 1.3025 0.15 20 2.6929 202 13.5 8.4 6.3 1.2450 0.2 20 8.550 814 18 13 10.6 1.22



Table 4 Crystal growth data for Tm-doped KLuW. A: Tm sub-stitution in solution relative to Lu [at. %], B: Cooling rate duringgrowth [K/h], C: Cooling interval [K], D: Crystal weight [g], E:Growth rate [10−4 g/h], G, H, and I: Crystal dimensions [mm]along the c, a*, and b crystallographic directions, respectively.

A B C D E G H I KTm0.5 0.2 20 3.9716 397 14.1 11.1 7.47 1.871 0.2 20 3.6277 362 14.3 10.11 5.4 1.17

3 0.1 20 5.6229 281 11.32 9.28 7.15 1.23

5 0.1 20 4.9860 249 15.2 10.4 7.5 0.997

24 25 26 273.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0 (a)

dist

ance

alo

ng N

p [m

m]

Yb content [at. %]

Np

Nm

0 1 2 3 428.0

27.5

27.0

26.5

26.0

25.5

25.0

24.5

24.0(b)

Yb

cont

ent [

at. %

]

distance along Nm [mm]

Figure 8 Yb distribution profile along the Np ≡ b (a) andNm (b) principal optical axes of KLu0.745Yb0.255W.

within the crystal. Typical sizes and weights of the growncrystals are about one cubic centimetre (see Fig. 9) andseveral grams. Tables 3 and 4 present data related to thegrowth parameters of Yb- and Tm-doped KLuW usingb-oriented seeds. Some attempts were performed also us-ing c-oriented seeds. Macro-defect free single crystals ofKLuW were obtained also in this case but the averagegrowth rate was slightly lower than for b-oriented seeds.

The morphology of the grown crystals is similar tothat we observed for other isostructural compounds ofthe same type [32,41,42], with the developed faces being{110}, {1̄11}, {010} and {310}, see Fig. 9. The externalmanifestation of the {010} face in these materials is im-portant as a structural reference in order to prepare samplesfor characterization purposes or laser applications. Usingthe same growth procedure we observed that the {010}face tends to decrease its area with increasing Ln atomicnumber which means that the growth rate of this face isincreasing [30,32,42].

According to the Hartman-Perdok theory [43], thecrystal morphology is governed by the presence of pe-riodic bond chains (PBCs) and we studied the morphol-ogy of KLuW from this point of view. Table 5 shows the

Figure 9 Morphology (a) and single crystals of undoped (b),5 at. % Yb-doped (c) and 3 at. % Tm-doped (d) KLuW.

Table 5 Strongest PBCs of monoclinic KLuW and their periods.

< uvw > PBC period [nm]1/2< 110 > 0.7315

< 001 > 0.74871/2< 112 > 0.7647

< 101 > 0.8042

Table 6 Observed faces {hkl} on the grown KLuW crystals,their interplanar dhkl-spacings, and the PBCs parallel to them.

{hkl} dhkl [nm] PBCs

110 0.6308 [0 0 1] [ 101 ] [ 211 ]111 0.6028 [1 0 1] [ 011 ] [ 211 ]

010 0.5107 [0 0 1] [1 0 1]021 0.3797 [ 211 ]

212 0.3669 [ 011 ]130 0.3140 [0 0 1]310 0.2586 [0 0 1]

four strongest PBCs which are simultaneously the PBCswith the shortest repetition periods in the KLuW structure.Table 6 lists the faces observed on the KLuW crystals ob-tained in our experiments and shows the PBCs parallel tothem. The growth rate of a {hkl} form is inversely pro-portional to the interplanar spacing dhkl, also included inTable 6. Faces with lower growth rate are more importantfor the morphology of the crystal because faces whichgrow faster tend to be less developed.



The interplanar dhkl–spacings, the shortest periods ofthe PBCs, and the unit cell parameters are affected bythe smaller radius of the Lu-ion and some trends can beobserved when comparing to other isostructural hosts likeKGdW: Thus, the ratio d010/d110 is somewhat smallerfor KLuW in comparison to KGdW. This ratio can beassociated with the external manifestation of the {010}face in competition with the {110} face – its decrease willbe expressed in the external habit of the grown crystalsto have smaller area of the {010} face and larger area ofthe {110} face. Furthermore, from the comparison of theunit cell parameters of the monoclinic potassium doubletungstates [1], it can be concluded that the relative decreaseof the b parameter when the atomic number of the Ln-element increases is more pronounced than that of the aand c parameters. This affects the tendency of developinga face which is proportional to dhkl. However, the slightmodification in the development of the {010} face could bealso related to the different physico-chemical properties ofthe solutions (e.g. KLuW-K2W2O7 vs. KGdW-K2W2O7).

The {110}, {1̄11} and {010} faces are parallel to twoor more PBCs; this agrees with the fact that these faces arewell developed, i.e. they grow slowly and the mechanismof the growth is slice to slice. The {130}, {310}, {021}and {2̄21} faces are parallel to one PBC and their growthrate is higher; hence, these faces are less developed and donot always appear. Figure 10 shows a projection parallel to[001] direction. One can see the interplanar spacing d110and the corresponding PBCs [001] which are linked in the[1̄10] direction. The interplanar spacing d020 correspondsto the form {010} which contains the [001] PBCs linkedin the [101] direction. Finally the interplanar spacings d130and d310 are associated with the forms {130} and {310},respectively; they also contain [001] PBCs but there are nolinks between the PBCs. The larger the interplanar spacingof a given form {hkl}, the higher the number of PBCsthat can be contained. As a consequence, the slice containsmore strong bonds, the rate of growth perpendicular to the{hkl} face is smaller and the face appears more developed.Furthermore, crystals grown from b-oriented seeds havethe [001] and [101] PBC directions as natural edges.

Besides some differences in the observed morphologyof the grown crystals, we note that the authors of [44],see also [45,46], favour c-oriented seeds for the TSSG ofKLuW because according to their experience such seedsprovide superior crystal quality and higher utilization ratio.

The production of thin doped layers is important forhighly absorbing materials, especially for the realizationof simplified (less pump passes) versions of the thin disklaser concept [47], the power scalability of which is in-versely proportional to the crystal thickness, but also ingeneral for a better overlap with the pump beam in the caseof diode pumping. The latter is relevant also to waveguidestructures which can be fabricated by different techniquesfor refractive index manipulation: e.g. oxygen ion im-plantation was recently used for planar waveguide struc-tures based on Yb:KLuW [48]. The monoclinic doubletungstates are very promising for thin film laser designs

Figure 10 Projection of the KLuW structure parallel to [001]showing the interplanar dhkl-spacings of the forms {010},{110}, {130} and {310} parallel to [001] PBCs.

Table 7 Face area mismatch for doped KLuW layers on KLuW.

epitaxial layer ƒ{010} ƒ{110} ƒ{310} ƒ{ 111 }

KLu0.88Yb0.12W 0.0739 0.0812 0.0637 0.0889KLu0.78Yb0.22W 0.1074 0.1306 0.1034 0.1347KLu0.48Yb0.52W 0.1436 0.2204 0.1570 0.2109KYbW 0.2156 0.3557 0.2556 0.3248*KLu0.97Tm0.03W 0.0689 0.0642 0.0568 0.0731*KLu0.95Tm0.05W 0.0859 0.0868 0.0772 0.0897*KLu0.925Tm0.075W 0.0855 0.0993 0.0776 0.1029*KLu0.90Tm0.10W 0.1161 0.1366 0.1041 0.1402*KLu0.80Tm0.20W 0.1984 0.2615 0.1957 0.2489

*for Tm-doping the composition refers to the solution

just because of the high doping levels possible and thelarge interaction cross sections. Thus, e.g. the stoichiomet-ric KYbW which exhibits only weak fluorescence quench-ing has an absorption length of only 13.3 µm for pumpingat 981 nm with polarization E//Nm [36].

We applied liquid phase epitaxy (LPE) for growth ofthin Yb- or Tm-doped KLuW crystalline layers on pas-sive KLuW substrates. KLuW is obviously the best choicefor Yb-doped monoclinic composites of this type becausethe lattice mismatch with respect to KYbW (0.33% onthe average for a, b, and c) is minimum [19]. Minimummismatch means reduced stress at the layer-substrate in-terface. Table 7 summarizes the area mismatch for thedifferent crystal faces defined by the expression f{hkl}= [SL{hkl}-SS{hkl}]/SS{hkl} where SS and SL are theareas calculated from the {hkl} periodicity vectors of thesubstrate and the layer (Yb- or Tm-doped), respectively.

For Yb:KLuW layers, the smallest lattice mismatchwhen increasing the doping level occurs for the {010}



face. For Tm:KLuW layers, the smallest lattice mismatchis for the {310} face, followed by the {010} face. Sincethe {010} face is simultaneously a principal optical plane,as described in the next section, it is still better to orientthe substrates parallel to this face.

The LPE growth technique is basically the same asfor bulk crystals but the special furnace has a wide zoneof uniform temperature to ensure almost vanishing tem-perature gradient. Also, the solvent part was increasedto 93–95 mol % for better control of the growth rate inaccordance with the solubility curves [19,24,49]. Afterhomogenisation of the solution, its saturation temperaturewas accurately determined with b-oriented KLuW seedsrotating at 40 rpm. The passive substrates were b-cut plateswhich were dipped into the solution with the [001] direc-tion perpendicular to the solution surface. Before insert-ing them into the furnace, the substrates were carefullycleaned, in rotation, in 1/1 HNO3/H2O (5 min), distilledwater (5 min), acetone (5 min), and ethanol (5 min). Inthe furnace, they were first heated for about 1 h abovethe surface of the solution and then dipped into the solu-tion for 5 min, at a temperature of 1 K above TS so thatthe surface part of the substrate is dissolved and cannotintroduce defects in the subsequent layer growth.

The epitaxial growth takes place for several hours ata temperature 1–6 K below the saturation temperature[19,49]. The 6 K difference corresponds to a supersatu-ration coefficient σ = 100(CG/CS − 1) ≈ 5.3% whereCG and CS are the solute concentrations at the growthand saturation temperatures. All LPE growth experimentswere performed with the crystal rotating at 40 rpm. Afterthe epitaxial growth, the composite crystals were removedfrom the flux and cooled slowly to room temperature toprevent any thermal stress, which could produce cracks.

The dependence of the LPE growth velocity on thecrystalline face was studied in [49]. It turns out that theepitaxial layer grows faster on the (010) face, Fig. 11a,which agrees with the fact that for bulk KLuW, the (010)face tends to be poorly developed, as mentioned above.The thickness of the epitaxial layer of KLu0.78Yb0.22Wgrown on a (010)-oriented KLuW substrate as a functionof time is shown in Fig. 11b: For short growth times (about6 h), the growth velocity is about 14.0 µm/h.

The doping level for the epitaxial layers was 5–50at. % for Yb [49] and 5 at. % for Tm [24]. Also for theepitaxial growth, the segregation coefficient of Yb and Tmwas close to or larger than unity. It tends to decrease whenthe Yb concentration in the solution increases (Table 8).The b-orientation of the epitaxial layer was verified by X-ray diffraction [24]. There was no appreciable diffusionof the dopant from the layer into the substrate (Figs. 11a,12).

The quality and the micromorphology of the grownepitaxial layers were studied by optical microscopy. Fig-ure 13 shows a confocal image and profile schemes ofthe micromorphology observed in KLu0.5Yb0.5W/KLuW(010). As can be seen, no polygonalization of the growthsteps occurs. The absence of polygonalization is due to

0 10 20 30 40 50 600.0

0.2

0.4

0.6

(110) (010) (310)

x in

KLu

1-xY

b xW

distance from the surface [µm]

(a)

0 5 10 15 20 25 300

50

100

150

200

250

300 (b)th

ickn

ess

[µm

]

time [h]

slope (6h) = 14 µm/h

slope (22h) = 7.1 µm/h

Figure 11 Measured Yb-density profile for a Yb:KLuW/KLuWcomposite on three different faces (a) (the inset shows the facesstudied), and growth velocity of a KLu0.78Yb0.22W epitaxiallayer on the (010) face of a KLuW substrate (b).

-20 0 20 40 60 80 100 120

0

1

2

3

4

substrate epitaxial layer

N=3.5±0.3x1020 at/cm3

Tm

3+ d

ensi

ty [1

020

at/c

m3 ]

distance from the interface [µm]

Figure 12 Tm-density distribution for the (010) face of aTm:KLuW/KLuW composite. The inset shows a microscopeimage of the layer region.



Figure 13 Confocal image andscheme of the two screw disloca-tions of opposite sign leading toa complex hillock growth pattern,observed in KLu0.5Yb0.5W/KLuW(010). The graphs depict the heightprofile of the growth steps in twodifferent regions of the growth mi-cromorphology.

Table 8 Segregation coefficient of Yb (KYb) and chemical for-mula of the KLu1−xYbxW film grown on KLuW, as functionsof the Yb concentration in the solution.

Yb content in the solution [at. %]

KYb chemical formula of the layer

5 1.24 KLu0.94Yb0.06W10 1.23 KLu0.88Yb0.12W20 1.11 KLu0.78Yb0.22W50 1.04 KLu0.48Yb0.52W

the fact that the process is governed mainly by diffusioneffects and not by structural effects related to the PBCs.The LPE growth is dominated by the superposition of twoscrew dislocations of opposite sign.

The distance between the growth steps amounts (inrelative units normalized by the radius of the dislocationsr∗) to L/r∗ ≈ 0.0125 for the step closest to the centre ofthe screw dislocations and increases with the separationfrom this centre. As can be seen from Fig. 13, the height ofthe growth steps strongly depends on the distance from thecentre. It amounts to ≈ 80 µm in the vicinity of the hillockcentre but as the separation increases, the height of thegrowth steps decreases, e.g. to ≈ 45 µm at L/r∗ ≈ 1.5.This effect is due to the splitting of the growth steps asseen in the marked area. Generally speaking, the epitaxialgrowth on the {010} faces exhibits mainly a flat surface,which indicates a layer-by-layer growth mechanism.

3. Properties of KLuW

The information previously available on the properties ofKLuW was limited to the average refractive indices at0.7 µm (≈ 2.08) [2] and 0.9. . . 1.4 µm (≈ 1.9) [8], thetransparency (0.3. . . 5–5.5 µm for a thickness of 1 mm)[2,8], the average thermal conductivity (≈ 3 W/m/K) [2],and the hardness in the Moh’s scale (4.5–5) [2]. We havecollected in Table 9 the results obtained since the “re-discovery” of KLuW as a laser host in 2004, see alsoTable 1.

The transmission of KLuW is shown in Fig. 14. For thismeasurement a 1-mm thick b-cut plate was used [16]. Atan absorption level of 1 cm−1 the transparency of KLuWextends from 365 to 5110 nm.

The monoclinic phase of KLuW belongs to the 2/mpoint group. Hence, KLuW is a biaxial crystal with inver-sion centre. The three orthogonal principal optical axes x,y, z are traditionally labelled for monoclinic crystals as Np,Nm, Ng . They are defined by the ratio of the correspond-ing refractive indices nx < ny < nz or np < nm < ng .In monoclinic crystals one of the principal optical axescoincides with the 2-fold symmetry axis (the crystallo-graphic b-axis). In the case of KLuW this is Np. Theother two principal optical axes lie in the a-c plane. Wedetermined their orientation at 632.8 nm with two crossedGlan-Taylor polarizers.

Figure 15 shows the orientation of the optical ellipsoidwith respect to the morphology and the crystallographicframe (both frames abc and NpNmNg are right-handed).The principal optical axis Ng is located at 18.5◦ with



0 1 2 3 4 5 6 7 80

20

40

60

80

tran

smis

sion

[%]

wavelength [µm]

Figure 14 Unpolarized transmission of undoped KLuW.

Figure 15 Optical ellipsoid of KLuW at room temperature.

respect to the c crystallographic axis and Nm is locatedat 59.2◦ with respect to the a crystallographic axis.

The dispersion curves for the three refractive indiceswere measured from 410 to 1200 nm by the minimum de-viation method using semiprisms [30]. Two prisms cut indifferent principal planes were used and np was mea-sured twice. The accuracy was 5 × 10−4. Figure 16shows the experimental values of the refractive indicesand the fitted curves using Sellmeier equations contain-ing a single UV pole and an IR correction term, n2 =A+B/[1−(C/λ)2]−Dλ2. The obtained Sellmeier coef-ficients, valid in the visible and near-IR, are summarizedin Table 10.

The angle 2Vg between the two optic axes of KLuWis defined from sin Vg = (ng/nm)(n2m − n2p)1/2(n2g −n2p)

−1/2. Following the existing conventions, these axesare in the Np-Ng plane and the bisectrix of 2Vg coin-cides with the Ng axis. Figure 17 shows the wavelengthdependence of 2Vg calculated from the measured refrac-tive indices. In the whole wavelength range KLuW is anoptically positive crystal (Vg < 45◦).

0.4 0.6 0.8 1.0 1.2

2.00

2.05

2.10

2.15

2.20

np

nm

ng

refr

activ

e In

dex

wavelength [µm]

Figure 16 Dispersion of the refractive indices of undopedKLuW at room temperature.

0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

82

84

86

88

90

92

2V

g [°

]

wavelength [µm]

Figure 17 Angle between the two optic axes of KLuW.

In order to determine the linear expansion coefficients,the unit cell parameters of KLuW were measured as afunction of the temperature by powder X-ray diffractionanalysis, at temperatures of 298, 323, 373, 473, 573, 673and 773 K [30]. This range of temperatures is sufficientlylarge to cover the variation of the unit cell parametersexpected under conditions of laser operation. As can beseen from Fig. 18, the unit cell parameters a, b and cincrease and the angle β remains basically constant withtemperature.

The linear thermal expansion coefficients in a givencrystallographic direction are α = (∆L/∆T )/LRT ,where LRT is the initial parameter at 298 K (room tem-perature), and ∆L is the modification of this parameterwhen the temperature is changed by ∆T . The linear ther-mal expansion coefficients are obtained as slopes of the(∆L/LRT ) dependence on the temperature T (Fig. 18).The values for monoclinic KLuW are α100 = 10.6(2) ×10−6, α010 = 3.35(2) × 10−6, α001 = 16.3(2) × 10−6,αc∗ = 15.1(1) × 10−6, and αV = 29.2(3) × 10−6 K−1.From these results one can derive α13 and, in this way, the



Table 9 Some physical properties of the monoclinic KLuW crystal host. Data from [40] and [50] refer to 5 at. % Yb-doped KLuW.

properties [Ref.]temperature of polymorphic transformation/melting [K] 1312+ / 1326+, 1330 [40]crystal structure (space group – point group) monoclinic centrosymmetric (C2/c≡ 62hC - 2/m) [30]site symmetry / coordination number / Lu3+ ionic radius C2 / 8 / 0.977 Å [30]lattice constants a=10.576(7) Å, b=10.214(7) Å, c=7.487(2) Å, β=130.68(4)º, Z=4 [30]cell volume and density 613.3(6) Å3 and 7.686 g/cm3 [30]cation density / minimum Lu-Lu separation 6.52×1021 cm-3 / 4.045(3) Å [30]transparency range (1 cm-1 level for 1 mm thickness) 365-5110 nm [20]refractive index @ 1 µm np=1.995, nm=2.030, ng=2.084 [30]optical ellipsoid orientation (632.8 nm) Np//b, ∠(a,Nm)=59.2°, ∠(c,Ng)=18.5° [30]angle between the two optic axes at 1064 nm 2Vg=82.03° (optically positive) [30]strongest phonon modes [cm-1] 908+, 756+ [3]ƒ

specific heat @ 300 K / 363 K / 1099 K 324.4+ / 350+ (365) / (701) J/kgK ([40])thermal conductivity coefficients κ @ 298 K / 563 Kand orientation of the conductivity ellipsoid, X1X2X3

κ1=3.09/2.49 W/mK, κ2=2.55/2.05 W/mK, κ3=4.40/3.15 W/mK X2//b, ∠(a,X1)=6.34°/1.94°, ∠(X3,c)=34.4°/38.8°

[40]**

thermal expansion coefficients α [10-6 K-1] andorientation of the thermal expansion ellipsoid, X1X2X3

α11=8.98(12.8), α22=3.35(7.8), α33=16.72(22.2) X2//b, ∠(a,X1)=27.24°(30.37°), ∠(X3,c)=13.44°(10.37°)

[30]*([50])**

microhardness numbers for planes ⊥ (axis) VH=440(a*), 410 (b), 560(c)VH=393.5

KHN=366 (a), 417 (b), 489 (c)

[30][11]***[40]

+ unpublished (this work); f infrared and Raman spectra for 5 at. % Yb-doped KLuW can be found in [37];* measurements of the unit cell parameters at 298–773 K by powder X-ray diffraction;** measurements by a thermal dilatometer at 323–873 K [50], other data for 5 at. % Yb and 6 at. % Tm co-doped KLuW can befound in [29];*** for 3 at. % Nd-doped KLuW with unspecified orientation.

References in additional brackets correspond to the values given in brackets.

300 400 500 600 700 8000.0

0.3

0.6

0.9

1.2

1.5

1.8abc

Vc cos( -90°)

(L-L

RT)

/LR

T x

102

temperature [K]

Figure 18 Evolution of the unit cell parameters of KLuW withtemperature (RT: room temperature).

linear thermal expansion tensor at 298 K in a crystallo-physical coordinate system with orthogonal axes parallelto a, b, and c*.

By diagonalizing the obtained matrix one can obtainthe linear thermal expansion tensor in the eigenframe X1,

Table 10 Sellmeier coefficients of KLuW.

A B C [µm] D [µm-2]

np 3.21749 0.75382 0.25066 0.05076nm 3.36989 0.74309 0.26193 0.04331ng 3.58334 0.73512 0.26700 0.02953

X2//b, X3. In monoclinic crystals the 2-fold symmetryaxis (the crystallographic b axis) coincides with one of theprincipal axes. In the case of KLuW, this axis correspondsto the minimum thermal expansion coefficient, i.e. X2.The diagonalized linear thermal expansion tensor has thefollowing values:

αij =

8.98 0 00 3.35 00 0 16.72

× 10−6 K−1

The principal axis with medium thermal expansion, X1,is located at ρ = 27.24◦ from the a axis and the principalaxis with the maximum thermal expansion coefficient, X3,is located at δ = 13.44◦ from the c axis. The thermalexpansion ellipsoid is depicted in Fig. 19.

Comparing with previous data obtained for KGdW,KYW, KErW and KYbW [51], KLuW has a similar an-



Figure 19 Thermal expansion ellipsoid of KLuW.

gle between the X3 and the c axes. On the other hand,if the ratio α33/α11 is considered as a measure of thelinear thermal expansion anisotropy in the face {010}one can conclude that this anisotropy decreases along theKLnW series. Thus, KLuW is the host with the lowest ther-mal anisotropy in the {010} plane. This lowest anisotropymeans reduced probability of cracking for thermal reasonswhen KLuW crystals are used as laser active elements.Finally, the calculated linear expansion coefficients alongthe principal optical axes Np, Nm, and Ng , amount to3.35×10−6, 11.19×10−6, and 14.55×10−6 K−1, respec-tively.

The specific heat of KLuW was measured by meansof the relaxation method over a temperature range fromthe pumped helium temperature (2.5 K) up to about 400 Kwhere it shows no anomaly, Fig. 20a. From the low tem-perature part of the data (phonon contribution), a Debyetemperature of θD ≈ 303 K was obtained, Fig. 20b.

Cutting and polishing are important prerequisites forsuccessful applications of any crystal material. The tech-niques for cutting and polishing are directly related to themechanical properties. High hardness values ensure betterquality of the polished surfaces of the active elements.

Vickers static micro-indentations have been performedon three different faces of KLuW. The faces were cut andpolished (to 0.3 µm metallographic quality) perpendicularto the three crystallographic directions a*, b and c, respec-tively. The used strengths were 0.03, 0.05 and 0.1 N. Thecorresponding loading rates were 0.01, 0.02 and 0.05 N/s,respectively, and each indentation was for 1 s. For eachload, different indentations were performed and the av-erage diagonal imprints were used for the calculation. Ascanning electron microscope was used to quantify thediagonal imprint of the residual indentation impression;the magnification was ×12000 [30].

The Vickers diamond pyramid hardness number (HV )is defined by the applied load divided by the surface areaof the indentation HV = αHV F /d2, where F is the appliedload in N, d is the arithmetic mean of the two diagonals,d1 and d2. In Vickers hardness number (1 HV = 9.8 MPa),units αHV = 1.89×105. Figure 21 shows a typical Vickersindentation on the (010) face of KLuW.

0 100 200 300 4000

100

200

300

400

spec

ific

heat

Cp

[J/k

g K

]

temperature T [K]

(a)

0 50 100 150 200 2500.0

0.1

0.2

0.3

D=303.45K

Cp/

T [J

/kg

K2 ]

T2 [K2]

(b)

Figure 20 Specific heat Cp of KLuW (a) and its lattice com-ponent Cp/T (b).

Figure 21 Vickers indentation on the (010) face of KLuW witha load of 0.05 N.

The calculated HV, perpendicular to the a*, b and cdirections of KLuW, is 440, 410 and 560 HV, respectively.Hence, this parameter exhibits pronounced anisotropy forKLuW. The Vickers value HV perpendicular to the b di-rection, transformed to the Moh’s hardness scale, gives avalue of about 4. Hence, the (010) face, being the softest,is not the proper choice from the point of view of efficientpolishing procedure.



900 950 1000 10500

1

2

900 950 1000 10500

5

10

15

900 950 1000 10500

1

2

E//Ng

E//Nm

cros

s se

ctio

ns [1

0-20

cm2 ]

abs

em

E//Np

wavelength [nm]

Figure 22 Polarized measured absorption and calculated emis-sion cross-sections of Yb:KLuW.

4. Spectroscopy of Yb- and Tm-doped KLuW

Polarized optical absorption measurements of Yb:KLuWand Tm:KLuW at room temperature were performed inorder to establish the best conditions for pumping of suchlasers and to estimate the absorption and emission crosssections using the reciprocity method [52]. Here we focusonly on the emission near 1 µm and near 1.9 µm, forthe 2F5/2 →2F7/2 and 3F4 →3H6 transitions of Yb3+and Tm3+, respectively, because these transitions werestudied in the laser experiments (next sections). Polarizedabsorption and luminescence studies at low temperature(6–10 K) were used to determine the Stark levels of Yb3+

and Tm3+ in KLuW. The main spectroscopic parametersfor the two dopants in bulk KLuW are summarized inTable 11.

Yb3+ and Tm3+ have odd 4f13 and even 4f12 numbersof active 4f electrons, respectively. Both dopants, substi-tuting Lu3+, occupy a C2 symmetry site. The interactionwith the crystal field results in splitting of the free ion2S+1LJ terms into (2J+1)/2 Stark sublevels for Yb3+ and(2J+1) Stark sublevels for Tm3+.

All samples used for spectroscopic characterizationwere cut and polished with their parallel faces normal to

900 950 1000 1050 11000

2

4

6

8

0

500

10000

10800

300 K 10 K

(0')

(

3)(0'

)

(2)

(0')

(

1)

(0')

(0

)

excitation

inte

nsity

[arb

. uni

ts]

wavelength [nm]

lase

r

pum

p

(2')(1')(0')

(3)(2)(1)(0)

ener

gy [c

m-1

]

Figure 23 Room temperature (blue line) and 10 K (green line)emission spectra recorded with 0.7 at. % Yb-doped KLuW.The inset shows the schematic diagram of the Stark levels andtransitions of Yb3+ in KLuW.

1000 1020 1040 1060 1080 11000.0

0.1

0.2

0.3

0.4

0.5=0.05=0.10=0.15=0.20=0.25

(a) E//Nm

gain [1

0-20

cm

2 ]

wavelength [nm]

1000 1020 1040 1060 1080 11000.0

0.1

0.2

0.3

0.4

0.5=0.05=0.10=0.15=0.20=0.25

(b) E//Np

gain [1

0-20

cm

2 ]

wavelength [nm]

Figure 24 Gain cross section σgain = βσem − (1 − β)σabs forpolarization along the Nm (a) and Np (b) axes of Yb:KLuWand different population inversion rates β.

one of the principal optical axes which allows to studythe other two polarization directions.



Table 11 Some spectroscopic properties of Yb- and Tm-doped monoclinic KLuW (from [15, 16, 22, 23, 33]).

spectral parameters Yb:KLuW Tm:KLuWsegregation coefficient / maximum doping level 1.3-1.4 / 100% 1-1.2 / 20%lower laser level [cm-1] 559 435 530thermal population @ 300 K [%] 4.22 7.64 1.942F5/2 fluorescence lifetime [µs] 275-299 (6.8 at. %) 254 (13 at. %)* -calculated 2F5/2 radiative lifetime [µs] 320 -3F4 fluorescence lifetime [ms] - 1.34 (3.7 at. %),

+ 0.9 (5 at. %)+

absorption wavelength for E//Nm and E//Np [nm] 981.1 980.9 802 793.5absorption linewidth for E//Nm and E//Np [nm] 3.6 4 4 1σabs for E//Nm and E//Np [10-20cm2] 11.8 1.8 5.95 9.96laser (reference) wavelength λref [nm] 1040 1950emission bandwidth (FWHM) [nm] ≈22 ≈28 - -σem [10-20 cm2] @ λref 1.01 1.24 1.20 0.57σreabs [10-20 cm2] @ λref 0.06 0.07 0.11 0.025* i h l h d ( hi k) + i h l h d

* pinhole method (this work), + pinhole method.

Table 12 Experimental Stark energy levels, Eexp, of Yb3+ and Tm3+ions observed in KLuW.

2S+1LJ Eexp [cm-1]

2F7/2 0, 175, 435, 5592F5/2 10187, 10498, 10735

2S+1LJ Eexp [cm-1]

3H6 0, 135, 155, 224, 247, 256, 279, 329, 346, 359, 513, 522, 5303F4 5663, 5711, 5724, 5768, 5876, 5963, 5976, 5981, 60023H5 8231, 8369, 8379, 8389, 8441, 8452, 8481, 8500, 8599, 8612, 86543H4 12603, 12606, 12717, 12729, 12744, 12790, 12801, 12883, 129063F3 14493, 14511, 14529, 14561, 14564, 14617, 146253F2 15078, 15081, 15103, 15123, --1G4 21092, 21121, 21128, 21233, 21353, 21361, 21535, 21581, 216131D2 27743, 27808, 27987, 28000, 28050

Figure 22 shows the measured absorption and calcu-lated emission cross sections for the single 2F7/2 ↔2F5/2transition of Yb3+ in KLuW and the three polariza-tion directions. The maximum absorption cross-section at981.1 nm calculated from the actual Yb3+ concentrationof 4.5×1019 cm−3 (0.7 at. % Yb-doped sample) amountsto 1.18×10−19 cm2 for E//Nm (linewidth: 3.6 nm). Bothvalues are very close to those reported for 5 at. % (inthe solution) Yb-doped KYW or KGdW [53,54] and thestoichiometric KYbW (100 at. % Yb) [36]. The maximumabsorption cross section for light polarization parallel tothe Nm principal optical axis is about 15 times larger thanthat ofYb:YAG [52]. The absorption profiles inYb:KLuW,for all polarizations, are very suitable for pumping withInGaAs laser diodes operating near 980 nm. The max-imum emission cross section in Yb:KLuW amounts to1.47×10−19 cm2, also for E//Nm at 981.1 nm.

Similar to Yb:KYW and Yb:KGdW, the useful polar-izations for Yb:KLuW are E//Nm and E//Np because theemission cross sections for E//Ng are very low. No differ-

ence was observed in the spectroscopy of thin Yb-dopedepitaxial layers [49].

The energy position of the four sublevels of the groundstate multiplet 2F7/2 and the three sublevels of the excitedstate multiplet 2F5/2 of Yb3+ in KLuW obtained from lowtemperature absorption and emission spectroscopy are in-cluded in Table 12. The Stark splitting of the 2F5/2 mul-tiplet is indicative of stronger crystal field in comparisonto Yb:KYW and Yb:KGdW [53,54] which is advanta-geous for tunable and short pulse laser operation. Fig-ure 23 shows emission spectra recorded at room and lowtemperature and the transitions involved in laser operation.

In the quasi-three-level laser system of Yb3+ the po-tential gain bandwidth for tunable or mode-locked opera-tion can be estimated by calculating the gain cross sectionwhich depends on the inversion rate. This is illustrated inFig. 24 for the two polarizations E//Nm and E//Np.

Measurements of the fluorescence lifetime performedwith a low doped, 0.7 at. % Yb:KLuW, sample to mini-mize the effect of radiation trapping yielded a decay curve



0.2 0.5 0.8 1.1 1.4 1.7 2.0250

300

350

400

450 (6.8 at.% Yb:KLuW)=(275±8) s (13 at. % Yb:KLuW)=(254±9) s

fluor

esce

nce

lifet

ime

[s]

pinhole diameter [mm]

(a)

0.2 0.5 0.8 1.1 1.4 1.7 2.0

240

270

300

330

360

390 (12 at. % Yb:KLuW/KLuW)=(241.1±25) s(52 at. % Yb:KLuW/KLuW)=(224.3±2.3) s

fluor

esce

nce

lifet

ime

[µs]

pinhole diameter [mm]

(b)

Figure 25 Fluorescence lifetime measurements using the pin-hole method of bulk (a) and epitaxial (b) Yb:KLuW.

that could be fitted by a single exponential correspondingto a time constant of 375 µs [16]. The use of the pinholemethod [55] which allows to eliminate this effect at higherdoping gave for different samples of 6.8 at. % Yb-dopedKLuW values of 299 and 275 µs. Figure 25 shows theresults of simultaneous measurements of two bulk andtwo (010) epitaxial samples which demonstrate the weakquenching effect in this material. Using the same proce-dure as in [36], i.e. the Füchtbauer-Ladenburg equation,an estimation of 320 µs was obtained for the radiative life-time of Yb:KLuW by averaging over the emission spectrafor the three polarizations. More data on the spectroscopyof Yb:KLuW can be found in [16].

The polarized room temperature absorption spectra ofTm3+ in KLuW were measured with a sample of 3.7at. % Tm-doping (Tm3+ density of 2.41×1020 cm−3).Figure 26 shows the absorption cross sections for the3H6 →3H4 transition, and the measured absorption andcalculated, by the reciprocity method, emission cross sec-tions for the 3H6 ↔3F4 transition of Tm3+ in KLuW.Strong anisotropy is characteristic for the KLuW host andthis is manifested also in the case of Tm-doping.

Similar to the case of Yb-doping, the useful polar-izations for Tm:KLuW are E//Nm and E//Np becausethe cross sections for E//Ng are very low. The maxi-

1600 1800 20000

2

4

1600 1800 20000

1

2

3

760 780 800 820 8400

3

6

9

E//Np

E//Nm

cros

s se

ctio

ns [1

0-20

cm

2 ]

abs

em

wavelength [nm]

abs

em

E//Ng

E//Nm

E//Np

abs

Figure 26 Polarized absorption cross sections of Tm:KLuWfor the 3H6 →3H4 transition, and absorption and emissioncross sections of Tm:KLuW for the 3H6 ↔3F4 transition andpolarizations E//Nm and E//Np.

1650 1800 1950 21000

2

4

63F

4

3H6

inte

nsity

[arb

. uni

ts]

wavelength [nm]

E//Nm

E//Np

Figure 27 Room temperature polarized emission spectra for the1.9 µm transition of Tm3+ in KLuW recorded with a 3.7 at.% Tm-doped sample.



0

5000

10000

15000

20000

25000

30000

1.4

m70

0 nm

65

0 nm

1.9

m

1G4

3F2+3F3

3H43H53F4

3H6

ener

gy [c

m-1]

CR

3

CR

4

CR

3C

R2

CR

2

CR

4

CR

1

CR

1

1G4

3F2+3F3

3H43H53F4

3H6

450

nm

480

nm

pum

p

Tm3+Tm3+Tm3+

1D2

1G4

3F2+3F3

3H43H53F4

3H6Figure 28 Energy diagram ofTm3+ in KLuW with arrows in-dicating radiative and non-radiativeprocesses; CR: cross relaxation.

mum absorption cross section for the 3H6 →3H4 transi-tion amounts to 5.95×10−20 cm2 for E//Nm at 802 nm.This main line with a width of 4 nm is very suitable fordiode pumping with AlGaAs laser diodes emitting near800 nm. The maximum absorption cross section for E//Npamounts to 9.96×10−20 cm2 at 793.5 nm but this Starkcomponent has a width of only 1 nm. The above valuesare similar to those known for Tm-doped KGdW [56],KYW [57], and KYbW [58].

The maximum emission cross section for the3F4 →3H6 transition amounts to 3.71×10−20 cm2 forE//Nm at 1841 nm. For polarization E//Np, the maximumemission cross section is 1.58×10−20 cm2 at 1820 nm.Room temperature luminescence spectra for this transitionwere recorded under 802 nm excitation, Fig. 27. Usingthe experimentally measured luminescence profiles andthe radiative lifetime calculated by the Judd-Ofelt method,the emission cross sections were calculated in [33] alsousing the Füchtbauer-Ladenburg equation: for E//Np theywere similar in magnitude to the results in Fig. 26 but forE//Nm – roughly 2 times lower.

The energy levels of Tm3+ in KLuW (Stark sublevels)were determined from low-temperature (6 K) absorptionand emission spectroscopy using a sample of 0.6 at. %doping [33]. The results are included in Table 12. Theupper level 3F4 for the 1.9 µm transition can be populatedby non-radiative relaxation and by four cross relaxationprocesses, as shown in Fig. 28.

The Stark splitting of a given manifold in Tm:KLuW isstronger than in Tm:KGdW. This again indicates stronger

crystal field in KLuW. In particular, the larger splitting ofthe ground state is favourable because this leads to lessthermal population of the lower level in quasi-four-levellaser operation. Further spectroscopic information on theother transitions can be found in [33].

Thulium lasers operate for the 1.9 µm transition as aquasi-four-level system and their spectral characteristicscan be predicted from the corresponding gain cross sec-tions (Fig. 29). Somewhat higher gain is predicted for theE//Nm polarization while for the E//Np polarization thelower wavelength limit seems slightly extended. Also, atlower inversion levels there is a trend that the supportedgain bandwidth is larger for the E//Np polarization.

In general the fluorescence lifetime values for the 3F4level, obtained using the pinhole method, are shorter thanprevious measurements of Tm:KYW and Tm:KGdW with-out taking into account the reabsorption effect. However,using the same method, the lifetimes measured in KLuWand KGdW were equal [23]. There is some dependence ofthe fluorescence lifetime of Tm:KLuW on the doping level(quenching): 1.34 ms and 0.9 ms were obtained using sam-ples of 3.7 and 5 at. % doping, Fig. 30 (see Table 4). Theradiative lifetime calculated using the Judd-Ofelt methodamounts to 1.483 ms [33].

5. Bulk and epitaxial Yb:KLuW lasers

The laser operation of the bulk and epitaxial Yb:KLuWsamples was studied in three- or four-mirror cavities



1800 1900 20000.0

0.4

0.8

1.2

1800 1900 20000.0

0.2

0.4

0.6 =0.1=0.2=0.3=0.4=0.5

(b)

wavelength [nm]

(a) =0.1=0.2=0.3=0.4=0.5

E//Nm

gain

[10-

20 c

m2 ]

wavelength [nm]

E//Np

Figure 29 Gain cross section σgain = βσem-(1-β)σabs forpolarization along the Nm (a) and Np (b) axes of Tm:KLuWand different population inversion rates β.

0.0 0.4 0.8 1.2 1.6 2.0

0.8

1.2

1.6

2.0

2.4 (3.7 at. % Tm:KLuW)=(1.34±0.07) ms(5 at. % Tm:KLuW)=(0.90±0.08) ms

fluor

esce

nce

lifet

ime

[ms]

pinhole diameter [mm] bulk

Figure 30 Dependence of the measured lifetime of the 3F4 levelof Tm3+ in KLuW on the pinhole diameter for 3.7 and 5 at.% doping levels. Bulk: no pinhole.

(a)M2M1

FL

Yb:KLuW/KLuW

M3

pump

output

(b)M2M1

FL

Yb:KLuW

M3 M4

pump

output

(c)

M2

M1

FL

Raman

Yb:KLuW fiber

fundamental

SA

M3

Figure 31 Astigmatically compensated three- (a) and four- (b)mirror cavities for longitudinal pumping of the Yb:KLuW laser.In (a), the radii of curvatures, RC, of M3 and M2 are −5 and−10 cm, respectively; in (b) the folding mirrors (M2-M3) haveRC = −10 cm; in both cases FL is a lens with f = 6.28 cm.Hemispherical cavity for power scaling (c): The output coupler(M2) has a RC = −2.5 cm, SA denotes saturable absorber, andM3 is a dichroic mirror separating the fundamental and Ramanradiation.

(Figs. 31a,b) with Ti:sapphire laser (1 nm linewidth)pumping or using a tapered diode laser (TDL) with 1 nmlinewidth and M2


0.5 1.0 1.5 2.00.0

0.2

0.4

0.6

0.8

1.0

(b)(a)

TOC

=2.8%, =49.6%

TOC

=5.0%, =51.7%

P=982nm

0.0 0.5 1.0 1.5 2.00.0

0.2

0.4

0.6

0.8

1.0

outp

ut p

ower

[W]

absorbed power [W]0.0 0.5 1.0 1.5 2.0

P=979nm

TOC

=2.8%, =47.3%

TOC

=5.0%, =55.75%

0.5 1.0 1.5 2.0absorbed power [W]

0 300 600 900 12000

30

60

90

120

150

180

P=980nm

outp

ut p

ower

[mW

]

absorbed power [mW]

TOC

=1.5%, =23.9%L=1047nm

TOC

=5.0%, =41.9%, L=1041nm

(c)

Figure 32 Output power ver-sus absorbed pump power ofthe Ti:sapphire laser pumpedYb:KLuW laser using the 140-cmlong cavity from Fig. 31b with dou-ble pass pumping (through 80%retroreflection of the residual pumpby the cavity mirrors) for two dif-ferent output couplers TOC : 6.8 at.% Yb-doping (a) and 13 at. % Yb-doping (b). Input-output character-istics obtained with single pass TDLpumping of the same 6.8 at. %Yb:KLuW sample in the same cav-ity from Fig. 31b (c). In all casesthe pump and laser polarizations areE//Nm.

an indication of the unaffected crystal quality and upperlevel lifetime by the increased doping. With an outputcoupler of transmission TOC = 2.8%, the Yb:KLuW laserthreshold could be reached for 965 nm < λP


Table 13 Maximum output power Pout and optical-to-opticalefficiency η0 achieved with bulk Yb:KLuW.

6.8 at. % Yb:KLuWPabs=1.90 W,λP=982 nm

13 at. % Yb:KLuWPabs=2.02 W,λP=979 nm

TOC[%]

Pout[mW]

η0[%]

λL[nm]

Pout[mW]

η0[%]

λL[nm]

1.5 465 24.5 1047.2 490 24.3 1051.32.8 800 42.1 1043.6 820 40.6 1049.05 910 47.9 1041.2 940 46.5 1046.5

10 965 50.8 1033.3 1010 50.0 1041.0

also attributed to the wavelength dependent reabsorptionlosses, see Fig. 24.

The imperfect overlap of the laser and pump modes inthe case of TDL pumping resulted in increased thresholdsand reduced slope efficiency η: At the optimum pumpwavelength of λP = 980 nm the recorded output powerversus absorbed power is shown in Fig. 32c. The maximumPout = 170 mW (TOC = 5%) for Pabs = 1.14 W gives anoptical-to-optical efficiency of η0 = 15%. The bleaching ofthe absorption in the case of Ti:sapphire laser pumping wassuppressed when the laser was aligned as a result of theincreased pump saturation intensity in the quasi-three-levelYb-system. In contrast, no bleaching was observed in thecase of TDL pumping. The almost unchanged oscillationwavelengths λL in Fig. 32c in comparison to Figs. 32a,band Table 13 indicate, however, that the net gain shouldbe almost the same.

The E//Np polarization was studied with an analogous6.8 at. % Yb-doped KLuW (3 mm thick) sample withNm-cut. For an absorbed power of 1.7 W (Ti:sapphirelaser pumping), an output power of Pout = 750 mW wasobtained at 1046 nm using an output coupler with TOC =5% (η = 54.2%) which is close to the 800 mW obtained forE//Nm with the 5 at. % Yb-doped b-cut sample (Fig. 32a).The thresholds with these two samples for TOC = 5% were280 and 250 mW, respectively.

Power scaling in the CW regime was studied using a3-mm thick, uncoated KLuW sample with 5.24 at. % Yb-doping and b-cut, mounted in a water-cooled Cu-block,and placed in the compact plano-concave cavity depictedin Fig. 31c, as close as possible (≈ 0.2 mm) to the planemirror M1. The pump waist in the position of the crystalwas 40 µm. The laser was optimized by adjusting the cav-ity length, operating near the hemispherical configuration.The oscillation wavelength was dependent on the outputcoupling, decreasing from λL = 1052.4 nm for TOC = 2%to λL = 1039.5 nm for TOC = 10%. The output beam waslinearly polarized with E//Nm independent of the outputcoupling.

Figure 33a shows the input-output characteristics forTOC = 2%, 5%, and 10%; output couplings of 0.5–2%led to very similar results. The laser configuration usedprecluded any measurement of the absorbed pump power

under lasing conditions because of the strongly divergentnature of the pump beam. Therefore, the output poweris plotted in Fig. 33a against the pump power incidenton the crystal. The absorbed power at threshold can bedetermined because the small-signal absorption remainsunchanged at low intracavity laser intensity. The absorbedpump power for reaching the laser threshold was 0.39,0.57, and 0.81 W, for TOC = 2%, 5%, and 10%, respec-tively.

The most efficient operation was realized applying theTOC = 2% output coupler. At an incident pump power of6.8 W, the output power reached 3.28 W leading to anoptical-to-optical efficiency of η0 = 48.2%; the slope ef-ficiency at high pump powers exceeding 4.0 W was η =78.2%. By moving the crystal away from the pump fo-cus, the small-signal absorption for the pump light wasestimated to be 85%–90%. Assuming that under lasingconditions the bleaching effect would be balanced by theincrease in the absorption due to the intracavity intensity,it is clear that the slope efficiency obtained with respectto the absorbed pump power is very close to the quantum-defect limit of 93%. For TOC = 5% and 10%, the highestoutput power obtained was Pout = 2.78 W and 2.61 W, cor-responding to slope efficiencies of η = 73.0% and 71.8%,respectively. The absence of an output power roll-off inFig. 33a at the highest pump level applied indicates apotential for further power scaling. The deviation of theinput-output characteristics from the linear law in Fig. 33a,i.e. the lower efficiency in the pump power range below4.0 W, is attributed mainly to the lower absorption (0.55–0.57 as compared to 0.6–0.72 for pump powers above4.5 W, measured without lasing). This pump absorptionvariation is caused by the wavelength shift of the diodelaser which emitted a multi-peaked spectrum: Its centrewavelength changed from 973 to 978 nm with increasingoutput power.

By introducing a Cr4+:YAG saturable absorber (a1 mm thick AR-coated plate with 80% transmission) intothe cavity (Fig. 31c), passively Q-switched laser opera-tion was achieved. Since SRS was expected to occur inthe nanosecond regime, the incoupling mirror M1 waschosen highly reflecting both at the laser and Ramanwavelengths. To stabilize the laser, the physical cavitylength was shortened from 25 to 22 mm, and in orderto avoid multi-pulse operation, a sufficiently large outputcoupling was required. The most appropriate output cou-pler available for this experiment had a transmission of10% at 1030 nm and about 20% near the Raman line at1140 nm. Slightly above the threshold (4.28 W of incidentpump power), SRS occurred and the output spectrum con-sisted of the fundamental line at 1030.6 nm and the 1st

Stokes line at 1137.6 nm which obviously corresponds tothe Raman-active vibration mode ν1, see Table 2. Figure33b shows the input-output characteristics of the passivelyQ-switched Yb:KLuW laser with simultaneous SRS self-conversion. At the highest available pump power (7 Wincident on the Yb:KLuW crystal), the total average out-put power reached 1.3 W. The part of the Raman radiation



0 1 2 3 4 5 6 70

1

2

3

outp

ut p

ower

[W]

incident pump power [W]

TOC

=2%, =78.2%, L=1052.4 nm

TOC

=5%, =73.0%, L=1045.2 nm

TOC

=10%, =71.8%, L=1039.5 nm

(a)

4.0 4.5 5.0 5.5 6.0 6.5 7.00.0

0.3

0.6

0.9

1.2

1.5

(b)

aver

age

outp

ut p

ower

[W]

incident pump power [W]

total output, =46.3% fundamental, =32.1% Raman, =14.1%

-4 -2 0 2 4 6 8 10 120

10

20

30

(c)

pow

er [k

W]

time [ns]

fundamental FWHM: 1.41 ns Raman FWHM: 0.71 ns

Figure 33 Output power versus absorbed pump power of theCW Yb:KLuW laser pumped by an unpolarized fiber-coupleddiode in the 2.5 cm long plano-concave cavity from Fig. 31c(a), average powers in the Q-switched regime (b), and temporalpulse profiles recorded at a repetition rate of 28 kHz (c).

was 0.4 W. The optical-to-optical efficiencies with respectto the incident pump power were η0 = 18.6%, 5.7%, and12.9%, for the total, the Raman, and the fundamental ra-diation, respectively. The corresponding slope efficiencieswere η = 46.3%, 14.1%, and 32.1%. It can be seen fromFig. 33b that, as in the case of CW operation, no roll-offoccurred up to the highest pump level applied, suggestingthat the laser can be further power scaled. The linear de-pendence of the output characteristics in Fig. 33b is dueto the high Q-switching threshold. This is consistent withthe CW operation results at higher pump powers.

For a passively Q-switched laser, the pulse repetitionfrequency depends on the pump power. In the present laserit increased almost linearly from 6 kHz at threshold to28 kHz at the highest pump level. The maximum energiesof the fundamental and Raman pulses at 28 kHz were 32.4and 14.4 µJ, respectively.

The temporal pulse profiles (see Fig. 33c) indicatepulse durations (FWHM) of 1.41 ns for the fundamentaland 0.71 ns for the Raman radiation. The pulse-to-pulsefluctuations for both the fundamental and the Raman ra-diation were estimated to be less than 5%. A peak powerof 23 kW for the fundamental pulse was calculated fromthe values of the pulse energy and the duration. For ac-curate estimation of the peak power of the Raman pulsewe took into account its exact temporal shape (Fig. 33c)and arrived at 15.2 kW.

In comparison to previous work on Q-switching per-formed with Yb:KYW and Yb:KGdW, the laser perfor-mance of Yb:KLuW is superior with respect to the pulseenergy, duration, and peak power [59–61]. In terms ofenergy and peak power the results are higher by morethan one order of magnitude than those reported in [59]and [60], and although comparable pulse durations weregiven in [61], the improvement in terms of energy andpeak power is still at least 3–4 times and 2 times, respec-tively. The efficiencies achieved with Yb:KLuW are alsohigher.

The CW laser experiments with an uncoated 12 at.% Yb:KLuW epitaxial sample (a 100 µm layer grown ona 1.1 mm thick undoped substrate) were performed inthe V-type cavity shown in Fig. 31a without special cool-ing. Again, the 1.2 mm thick b-cut Yb:KLuW/KLuW(010)crystal was placed under Brewster angle between the twofocusing mirrors. The sample was oriented for propaga-tion along the b(Np) axis with faces parallel to the Nm-Ng plane and polarization along the Nm principal opticalaxis. The CW input-output characteristics are presented inFig. 34a. We limited the incident pump power applied to1.85 W. The maximum output power of Pout = 415 mWmeasured corresponds to an optical-to-optical efficiency ofη0 = 55% with respect to the absorbed pump power (TOC= 3%). The slope efficiency with respect to the absorbedpower increases with TOC reaching a maximum value ofη = 66% (TOC = 10%, Fig. 34a). The efficiencies exceedthose achieved with the 2.2-mm thick 13 at. % Yb-dopedbulk KLuW in the similar pump and laser configurationfrom Fig. 31b. This is attributed to the strongly reduced



reabsorption which leads to about 4 times lower thresh-olds in the case of the epitaxial sample (about 70 mW inFig. 34a) and to shorter laser wavelengths λL. The rela-tively short oscillation wavelength λL (Fig. 34a) can bealso attributed to the stronger bleaching of the absorption(Fig. 35).

The dependence of the actual absorption (the low-signal absorption of this epitaxy was about 64%) on theoutput coupling is also rather pronounced (Fig. 35), justas a consequence of the strong absorption bleaching whenlasing is interrupted. In the case of Ti:sapphire laser pump-ing the absorption is bleached at relatively low incidentpowers (in this range it cannot be reliably estimated) andthe further dependence on the incident pump power israther weak.

We examined the influence of thermal effects by em-ploying a chopper with a 1:10 duty cycle. We observedonly a weak effect of about 10% at the maximum appliedpump powers (see Fig. 34a), i.e. the maximum averageoutput power achieved with the chopper was 45 mW.

Recycling of the pump using another end mirror M3in Fig. 31a, it was possible to increase the output powerin the CW regime to Pout = 515 mW (TOC = 3%) atλL = 1030 nm as a result of the increased absorption inthe two pump passes [19]. It is interesting to note thateven without cooling no damage of the epitaxial crystaloccurred regardless of the high power levels (intracavityintensity exceeding 1 MW/cm2 with the TOC = 1.1%output coupler) in the case of double pass pumping.

Using the TDL for single pass pumping in the sameset-up, CW laser operation was obtained for output cou-pler transmission between 1.1% and 10%, Fig. 34b. Alsoin this case, the laser thresholds obtained were lower (3 to5 times) in comparison to the bulk Yb:KLuW sample, e.g.as low as Pabs ≈ 120 mW for TOC = 1.1%. This mightbe related to the improved overlap of the pump and lasermodes in the thin epitaxial layer. At the maximum appliedpump power (1.25 W incident on the sample) the maxi-mum output power amounted to Pout = 105 mW and theoptical-to-optical efficiency with respect to the absorbedpower reached η0 = 20% (TOC = 3%). The highest slopeefficiency with respect to the absorbed power, η = 37.1%,was achieved with the TOC = 5% output coupler, Fig. 34b.

The actual absorption depends on the bleaching effectand, as already mentioned, can be substantially lower thanthe small-signal value. It depends, however, also on theoutput coupler transmission since the different intracavityintensity affects the pump saturation intensity, counter-acting the bleaching. Figure 36a shows the absorptiondependence on the incident pump power for three out-put couplers and also without lasing. The incident pumpintensity is comparable to or exceeding the saturation in-tensity and since the active layer thickness is smaller thanthe absorption length, the effect of absorption bleachingcan be clearly observed (Fig. 36a). Note that this was notthe case when pumping thick bulk samples of Yb:KLuWwith the same TDL. It can be seen from Fig. 36a that thebleaching effect is strongest when lasing is interrupted (in

0 200 400 600 8000

100

200

300

400

500

outp

ut p

ower

[mW

]

absorbed power [mW]

TOC

=1.1%, ,L=1040 nm

TOC

=3%, ,L=1032 nm

TOC

=5%, ,L=1029 nm

TOC

=10%, ,L=1026 nm

(a)P=981.5 nm

0 100 200 300 400 500 6000

20

40

60

80

100

120

P=981.3 nm

outp

ut p

ower

[mW

]

absorbed power [mW]

TOC

=1.1%, =20.3%, L=1030 nm

TOC

=3%, =31.2%, L=1030 nm

TOC

=5%, =37.1%, L=1030 nm

(b)

0 10 20 30 40 50 60 700

5

10

15

20

P=981 nm

outp

ut p

ower

[mW

]

absorbed power [mW]

TOC

=1%, =42%, L=1032 nm

duty cycle: 10%

(c)

Figure 34 Output power versus absorbed pump power for a 12at. % Yb:KLuW epitaxy pumped in the 67 cm long cavity fromFig. 31a by the Ti:sapphire laser (a), and by the TDL (b), andquasi-CW laser performance of a 52 at. % Yb:KLuW epitaxywith Ti:sapphire laser pumping: average powers (c).



0 300 600 900 1200 1500 18000.0

0.1

0.2

0.3

0.4

0.5

0.6

abso

rptio

n

incident pump power [mW]

TOC

=1.1%

TOC

=3%

TOC

=5%

TOC

=10%

nonlasing

Figure 35 Estimated single pass absorption of theYb:KLuW/KLuW(010) epitaxy in the case of Ti:sapphirelaser pumping versus incident pump power.

the M1-M2 arm in Fig. 31a). Lower transmission of theoutput coupler corresponds to increased intracavity inten-sity and consequently the absorption increases (Fig. 36b).But for a given TOC the dependence on the incident pumppower is not strong because the intracavity intensity alsoincreases with increasing pump power (Fig. 36a,b).

We tested also a 52 at. % Yb:KLuW epitaxy whichwas cut for the same propagation and polarization direc-tions but the doped layer was only 38 µm thick, Fig. 34c.However, linear dependence of the input-output character-istics was obtained only using a chopper. The maximumaverage output power obtained was Pout = 20 mW foran absorbed pump power of Pabs ≈ 63 mW which givesan optical-to-optical efficiency of η0 ≈ 32%. The averageoutput power obtained with the same output coupler ofTOC = 1% increased to 43 mW when double-pass pump-ing was applied. Although the absorption of this samplewas quite optimum, strong thermal effects made it verydifficult to achieve true CW operation. Pumping again inthe absorption maximum, in a single pass, only 17 mWof average output power at 1046.1 nm were obtained inthis case for an absorbed pump power of Pabs = 250 mW,using a special output coupler with TOC = 0.1%. Forincident powers exceeding 400 mW the output quicklydropped to zero and even optical damage was observedabove 500 mW. The extensive heating at high doping lev-els is attributed to impurities absorbing the laser and/orpump radiation which then relax non-radiatively. The in-fluence of excitation migration to impurities can be ruledout because no substantial shortening of the fluorescencelifetime was observed for the 52 at. % Yb-doped KLuWepitaxy (Fig. 25b).

Figure 37 shows the Z-shaped cavity used for mode-locking of the Yb:KLuW lasers. Two dispersion compen-sating prisms could be inserted in the arm with the out-put coupler. The semiconductor saturable absorber mir-

800 900 1000 1100 1200 13000.30

0.35

0.40

0.45

0.50(a)

abso

rptio

n

incident pump power [mW]

nonlasingT

OC=1.1%

TOC

=3%

TOC

=5%

0 1 2 3 4 5 6 7 8 9 100.30

0.35

0.40

0.45

0.50(b)

abso

rptio

n

intracavity power [W]

TOC

=1.1%

TOC

=3%

TOC

=5%

Figure 36 Estimated single pass absorption in the case of TDLpumping of the Yb:KLuW/KLuW(010) epitaxy versus the in-cident pump power (a), and versus the intracavity laser power(b).

ror (SESAM) was grown by the MOCVD-method andconsisted of a bottom Bragg mirror comprising 25-pairsAlAs/GaAs quarterwave layers designed for a centralwavelength of 1030 nm. Its reflection band extended from980 to 1070 nm. The 10-nm-thick InGaAs surface quantumwell structure had a saturable loss of ≈ 1%. Its relaxationtime was measured by the pump and probe technique tobe less than 5 ps.

The same 2.8-mm thick Yb:KLuW crystal oriented forE//Nm and the 3-mm thick crystal oriented for E//Np fromthe CW laser experiments, both with 6.8 at. % Yb-doping,were first compared with Ti:sapphire laser pumping, with-out special cooling. Without intracavity prisms, the laseroperated in the picosecond regime with a pulse repetitionrate of 98 MHz. Pulses as short as 2.8 ps near 1043 nmwere obtained for E//Np at a maximum average outputpower of Pout = 540 mW (TOC = 5%) corresponding to anoptical-to-optical efficiency of η0 = 32%. The measuredautocorrelation traces were fitted assuming a sech2-pulseshape, Fig. 38. The 13 nm broad (FWHM) spectrum in-dicates that the pulse duration in the picosecond mode



M2M1

FLYb:KLuW

M3 M4

pump

ps output

fs output

SESAM

P1

P2

M5

Figure 37 Schematic of the astigmatically compensated cavityof the mode-locked Yb:KLuW laser. The folding mirrors (M1-M3) are with RC = −10 cm, P1 and P2 are SF10 prisms, andFL is a lens with f = 6.28 cm.

-10 -5 0 5 10 150.0

0.5

1.0

1020 1040 10600.0

0.5

1.0p =2.8ps

p =30

inte

nsity

[arb

. uni

ts]

time delay [ps]

13 nm

wavelength [nm]

Figure 38 Autocorrrelation trace and spectrum (inset) of theYb:KLuW laser in the picosecond regime (Ti:sapphire laserpumping). τp: pulse duration (FWHM).

exceeds the Fourier limit by a factor of >30. The per-formance of the b-cut Yb:KLuW sample for E//Nm wassimilar in the picosecond regime.

The two SF10 Brewster prisms introduced for fem-tosecond operation had a tip-to-tip separation of 38 cm.The resulting pulse repetition rate was 95 MHz. TheFWHM of the shortest pulse obtained for E//Nm was81 fs (Fig. 39a), at an average output power of Pout =70 mW (TOC = 3%). The corresponding spectrum wascentered at 1046 nm and had a FWHM of 14.3 nm (in-set Fig. 39a). This results in a time-bandwidth productof 0.318 corresponding to transform-limited sech2-pulses.The results achieved for E//Nm and E//Np are comparedin Fig. 39b. Only minor differences can be seen, in ac-cordance with the similar bandwidths of the calculatedgain cross sections, Fig. 24. For both polarization orienta-

tions mode-locking was achieved for output couplers withtransmission TOC = 1...5%. The shortest pulse durationobtained for E//Np was 83 fs at λL = 1049 nm, for anoutput power of Pout = 36 mW (TOC = 1%). Substantiallyhigher output power, Pout = 295 mW, was obtained for apulse duration of 100 fs (Fig. 39b).

Using the TDL for pumping and the E//Np-orientedYb:KLuW crystal, stable femtosecond mode-locking wasachieved for output couplers with TOC = 1. . . 3%. A max-imum output power of Pout = 56 mW was obtained inthe mode-locked regime with TOC = 3%, for an inci-dent pump power of 1 W. The minimum pulse durationof 117 fs (FWHM) and the corresponding spectrum cen-tered at 1053 nm (Fig. 39c) give a time-bandwidth productof 0.39, hence the generated pulses are almost Fourier-limited.

Mode-locking with the 12 at. % Yb:KLuW epitaxywas realized in the same cavity (Fig. 37) with Ti:sapphirelaser pumping. In the picosecond regime, the laser oper-ated at a repetition rate of 100 MHz, the emission spectrumwas centered at 1030 nm and the FWHM of the gener-ated pulses was 1.8 ps (Fig. 40a) assuming sech2-pu

Date post:	25-Jan-2021
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

GrowthandpropertiesofKLu(WO ,andnovelytterbium ... · 2007. 5. 15. · Laser & Photon. Rev. 1,...

Documents